A computational model of auditory segregation of two frequency components—evaluation and integration of multiple cues
Kunio Kashino, Hidehiko Tanaka · Electronics and Communications in Japan (Part III Fundamental Electronic Science) · 1994
Abstract As part of the basic research on perceptual sound source separation, a quantitative model is presented which gives a probability of auditory segregation of two frequency components. The model consists of independent evaluation of multiple spectral features and integration of those derived values. First, psychoacoustic experiments which measure the probability of auditory segregation of two frequency components are described and evaluation functions of spectral features to approximate those experimental results are introduced. Then a method for integrating the spectral features is discussed and correlations between the model output and the results of the psychoacoustic experiments are measured.